Get up and running with Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.
ml/backend/ggml: set cpu n_threads
Michael Yangbf920883d54e
diff --git a/ml/backend/ggml/ggml.go b/ml/backend/ggml/ggml.goindex a55d42f14..f60619454 100644--- a/ml/backend/ggml/ggml.go+++ b/ml/backend/ggml/ggml.go@@ -301,6 +301,10 @@ func New(r *os.File, params ml.BackendParams) (ml.Backend, error) {bufts = append(bufts, bt)slog.Info("compute graph", "backend", C.GoString(C.ggml_backend_name(b)), "buffer_type", C.GoString(C.ggml_backend_buft_name(bt)))++ if C.ggml_backend_is_cpu(b) {+ C.ggml_backend_cpu_set_n_threads(b, C.int(params.NumThreads))+ }}return &Backend{